Measuring Availability the Player Focused Way: How Riot Games Changed Its...
summary
Riot defines availability as percentage of served player minutes using CCU data and incident records, builds ETL pipelines and Tableau heatmaps to roll incidents into six player-impact buckets (login, matchmaking, gameplay, store/payments, end‑of‑game, chat/inventory), sets priority one = >50% players impacted, chooses a 99% target, measures League of Legends at 98.97%, and uses monthly reports plus postmortems to drive actions that raise morale from ~1–2 to ~4.7, fund SRE hires, and grow live‑ops from ~30 to ~80–90 people.
AI-written summary · about
chapters (10)
- 0:00Riot's availability crisis
- 3:10Defining the problem
- 6:46Shifting to player-centricity
- 10:59Prioritizing incidents
- 13:32Establishing common language
- 17:30Designing the SLO
- 20:18Implementing the calculation
- 24:32Results and reporting
- 27:42Achieving culture change
- 33:34Key ingredients for success
related talks
49mWhat It Really Means for Your System to Be Available
25mParity in Observability Across Different Platforms
26mRiot Games: Evolution of Observability at the Gaming Company
32mTaming the Cost of Telemetry: How Riot Games Reined In...
26mUser Uptime in Practice
45mQuantifying Empathy with Service Level Objectives
31mNo-Nonsense Observability Improvement
20m9 SLIs; Oh My!